Evidentia

Inspiration

In today's digital world, information is everywhere, but finding reliable, accurate, and relevant information is still a major challenge. Students, researchers, developers, journalists, and professionals often spend hours searching through websites, academic papers, news articles, and reports to understand a topic and verify its facts.

Although AI chatbots can generate answers within seconds, those answers do not always provide sufficient evidence, traceable references, or a clear explanation of where the information originated.

What it does

Evidentia is an AI-powered autonomous research platform that transforms questions into evidence-backed research reports. It breaks a topic into smaller research questions, searches academic papers, trusted websites, and news sources, collects relevant information, and verifies claims by comparing evidence from multiple sources. It also identifies conflicting information, unsupported claims, and research gaps. Finally, it generates structured reports with traceable citations, helping students, researchers, and professionals find and understand reliable information more efficiently.

How we built it

We designed Evidentia as an AI-powered research platform using a React-based frontend and a Python backend with FastAPI. We planned an agent-based architecture where specialized AI agents handle research planning, source discovery, evidence extraction, claim verification, and report generation. We use AI inference services to process research tasks and external search and academic APIs to collect information from reliable sources. Supabase is planned for storing research sessions, sources, and extracted evidence. The collected information is organized into a structured evidence system and used to generate research reports with traceable citations.

Challenges we ran into

One of our main challenges was designing a system that could collect reliable information from multiple sources and verify it accurately. Coordinating different AI agents, handling conflicting information, maintaining citation accuracy, and reducing AI-generated misinformation required careful planning. We also had to consider API integration, response time, and efficient data management. These challenges helped us understand the complexity of building an autonomous AI research platform and encouraged us to focus on accuracy, transparency, and reliability.

Accomplishments that we're proud of

We are proud of creating Evidentia, an AI-powered research platform designed to go beyond ordinary AI-generated answers. Our idea focuses on collecting real information, verifying claims, identifying conflicting evidence, and generating research reports with traceable sources. We are also proud of designing a multi-agent workflow that divides complex research tasks into smaller, manageable steps.

What we learned

While developing Evidentia, we are learning about Agentic AI, multi-agent collaboration, AI orchestration, API integration, and evidence-based verification. We are understanding how different AI agents can work together to research topics, collect information, verify claims, and generate structured reports. Our biggest learning is that building reliable AI systems requires more than a powerful language model; it requires trustworthy sources, proper validation, and transparency.

What's next for Evidentia

Our future goal is to make Evidentia a reliable AI-powered research platform. We plan to integrate more academic and trusted information sources, improve evidence verification, detect research gaps, and introduce different research modes for students, researchers, journalists, and professionals. We also aim to improve report generation, citation management, and PDF exports to make research easier, faster, and more transparent.

Built With

• agentic-ai
• ai-agents
• evidence-verification
• fastapi
• groq
• llm
• multi-agent-systems
• natural-language-processing
• next.js
• openalex
• pdf
• postgresql
• python
• react
• report-generation
• rest-api
• semantic-scholar
• supabase
• tailwind-css
• tavily
• typescript
• web-search